Show HN: HN Resume to Jobs – AI Powered Job Matching Tailored to Your Resume (hnresumetojobs.com)
Hey HN! I'm excited to show off this side project I've been working on. This project matches your resume with the best matching jobs from the monthly HN Who's Hiring post. It works by creating a vector embedding of your resume using OpenAI's embedding API, and then ranking the jobs using a vector similarity score. (You can toggle between max inner product, cosine, and euclidean in the "Advanced Options")
I was laid off in August and it took a whole 6 months for me to find my new job. Fortunately, I found my new role on January's HN Who's Hiring post. So I hope this will prove useful to any job seekers out there. I know it's a tough time right now, but you will get through it!
Thanks HN! I would greatly appreciate any and all feedback!
82 comments
[ 5.5 ms ] story [ 156 ms ] threadIf you have a backend, perhaps add a feedback loop from the user which ones are relevant and which ones are not.
Using OpenAI, perhaps you can analyze the attributes and allow user to filter by Location, Title, size of the team, tags etc.
I actually do have a "Apply Filters" button, have you given that a try? I do thing the team size tag is a great idea.
I don't think the user of the post is the most important of the post, so I would use another color to not make it pop up.
My skill set is (css,) vanilla front end js and enough php to dumb db queries down the tube into the front end. There is no market for my write once works forever. I'm not interested in learning anything else but it would be wild to see an ordered list of suggestions accounting for marketability and difficulty. No way a puny human like myself can make sense of an n dimensional universe.
Every job I've ever taken, there were bullet points in the requirements that I didn't have. And every job has been in a different industry. (Granted, this is only 5 job hops in over 15 years) I'm not really sure I'd want to be perfectly qualified for a new job. Part of the fun is learning new stuff.
Of course, I do understand that any job is better than no job, and it's much easier to land one where you do hit all of the bullet points.
From there you could allow it to remix the skills while maintaining an average level of confidence greater than some number: so it'd be allowed to include things slightly out of your wheelhouse only if most of the job skills matched up perfectly.
You wouldn't need the math to check out perfectly there, the score would just be a heuristic it could use internally
These papers suggest that AI models could be employed to analyze your existing resume and the job specification, and then generate a tailored resume that highlights the most relevant skills and experiences. The main challenge lies in striking a balance between tailoring the resume and maintaining an accurate representation of your abilities.
[1] A Few-shot Approach to Resume Information Extraction via Prompts - 2022: https://arxiv.org/search/?query=A+Few+shot+Approach+to+Resum...
[2] Degendering Resumes for Fair Algorithmic Resume Screening - 2022: https://arxiv.org/search/?query=Degendering+Resumes+for+Fair...
recent advances would allow a platform to conversationally review your CV and probe you're desires to help with recommendations. super interesting space.
This. And you're like most people.
Most job descriptions and requirements lists are describing the "dream candidate". Very few employers actually expect anyone to have every requirement in the list.
Good luck
Other than that, it's like all good things - 80% of it is the button click, while the last 20% is proofing the final result and correcting the model so the machine can learn.
God Speed man. Love new tools.
Says who?
I find it actually quite strange to think of not only your own, but everyone else's professional (and implicitly: social, geographic) history as "inherently public" information.
I've only gotten jobs through people I've known since the 1980s but at least a capsule version on my resume is embedded in bios on countless event sites. It's hard (and undesirable) for a lot of people to keep that information hidden.
I would like to give this a try (been job hunting for the past 6 months), but I'm also a bit paranoid in sharing a lot of data, especially with a service that has no privacy policy.
As it stands, I am using google analytics, and also storing the resume text (though not the document), and the embedding of the text. I hope at the very least that provides some transparency.
Using Google Analytics might have some GDPR implications in Europe. A more GDPR-friendly alternative might be Plausible Analytics.
Also, given the amount of information contained in a CV, even without a name, address, contact info etc, it could make the process of identifying the actual person fairly easy, thus it might have some GDPR implications.
You could have a look at some Privacy Policies from recruiting agencies [0][1], though those are (in my opinion) fairly "abusive".
my 2 cents: given this is a side project, I would not store any data at all
[0]: https://www.greenhouse.com/de/privacy-policy
[1]: https://join.com/privacy-policy
The ICO has some good guidance. They’re your regulator in the U.K.
Use Wide Angle Analytics instead of Plausible or Google Analytics. Much better choice if you ever leak personal data to your analytics platform and very few solutions support PD. Wide Angle does.
Given the personal data you are processing, some might even become sensitive (disability status), you need DPO ( Data Protection Officer). There are DPO as a service out there.
I entered comma-separated list of keywords I'd like the job post to have and got quite nice results.
But congratulations on your new job!
Users will tell if it's useful, but regardless, I'm curious to apps that use LLMs in such ways... Returning lists.
I'm Operations so not a huge number of roles on this site though. So it might have effectively been giving me everything with "SRE" or "devops" in it.
Now, I wouldn't call my formal qualifications exceptional for the tasks I'm doing right now. You both need intimate tech knowledge and organizational skills, on top of some very specific knowledge about systems, algorithms, and AI. The width of tech you need to know for this job is kind of extreme, and IMHO that's not easy to read from my formal resume which mostly shows a background from teaching.
I do however have a personal tech blog and some projects on GitHub. I wonder how well their AI tracks such extraneous links, because if I was a tech company I'd first and foremost want to know specifics about what I know, how I apply that knowledge, and how easily I am able to learn and pick up new technologies. I think an AI is the perfect tool to automate such an analysis, so I'm wondering if that's what they did.
I guess opening Pandoras box would be to also employ forensic text fingerprinting to find other projects from the same author under various pseudonyms.
Are the jobs up to date? How are you sourcing jobs? Are you searching indeed.com?
* I live in Germany, and want to work remotely.
Then it filters out all the onsite jobs as well as the remote US or remote UK only jobs. It should be smart enough to know that Germany is in Europe, so Europe or anywhere remote jobs should remain.
Then I would like if it could extract some features and show them for the remaining jobs, and ideally I could filter by those features.
Features I'm interested in:
* type of employer (government, corporation, startup, non-profit)
* ethical orientation (wouldn't want to work for an ad company that sell user data, would very much like to work for an EV charging startup that tries to contribute to solving climate change). If that doesn't work, industry might serve as a proxy
* technologies used
* role type (developer, architect, product owner etc.)
* salary range (if available)
As for EV related jobs, what I have been doing is checking as many related conferences as I could find and EV related channels in YouTube and visit each one of the homepages for the companies. For conferences it can take up to 4h, but it's very quick (but tedious) to remove chaff that you're not interest into.
I'm haven't been lucky, not many are in France, very few are remote, but I have seen quite many in Germany, so maybe you're in better luck.
Also, https://climatebase.org/ and maybe another reread of https://news.ycombinator.com/item?id=26301765
One thing that would make it a lot more useful IMO would be if the posts were somehow collapsed at first, showing the headline information about the role, so I could open the ones that look interesting or sound like a good fit instead of needing to scroll past their full text.
I don’t think I see many attorney listings in those threads.